Some speech recognition systems use a pronunciation dictionary or other pronunciation information to recognize speech. Often, a pronunciation dictionary is constructed manually by linguistic experts, which can be time consuming and costly.